Notes for meeting of Monday, 13 January 2020. 10:00 CET / UTC 9:00 on #blender-coders on blender.chat.
Blender 2.82
As part of maintenance ensure 2.81 and 2.81a are properly tagged. After that the old 2.81 stabilizing branch can be dropped.
Linux archive name will be simplified to blender-$version$-linux64.tar.gz , dropping the modifiers glibc217 and x86_64 .
Sergey Sharybin will see to it Mantaflow gets fixed by Sebastián Barschkis.
Blender 2.83
Antonio Vazquez is working on Grease Pencil refactor. As part of this is refactor of area related to datablock evaluation for modifiers. He requests help from Sergey Sharybin on this.
LANPR will be after 2.83
Other
Dalai Felinto would like to move the morning meeting from 10:00 CET to 11:00 CET. Monday morning meetings will from 27th January on be on 11:00 CET (10:00 UTC).
Changes / New Features
Cycles
Add OptiX AI denoiser support ( Patrick Mours )
GPencil
Remove is_annotation parameter ( Antonio Vazquez )
Gizmo
Support 3D arrow gizmos in 2D spaces ( Campbell Barton )
Use compatible scale values between 2D/3D gizmos ( Campbell Barton )
Use pivot center for UV gizmos ( Campbell Barton )
Keymap
Industry Compat keymap: Support MMB for immediate transform tool use again ( William Reynish )
Sculpt
Pose Brush with Inverse Kinematics ( Pablo Dobarro )
USD Exporter
Properly export objects whose parents are not exported ( Sybren A. Stüvel )
This is great, although I’ve got a scene featuring DOF that shows heavy compression-like artifacts when using the Optix Denoiser options other than only Color.
Haven’t tested DOF and Optix Denoiser options outside the scene. Do others have the same issue?
Edit 2: There is more to it. I’ve changed colors, and artifacts was different for different color patterns. I can’t see any truths to it though. It’s like optix don’t like blue/green shades but… not always.
I am observing how devs are quickly organizing after Tracker curfew , they are now much faster at sorting bugs by categories, they are transforming random proposals into to-do and know issues, and everything is accelerating.
it is fascinating, I am also learning many things even if i am only a passionate observer and at the most I report some bugs or I give some suggestion when it happens.
I am happy to be here in this historical moment and to have the opportunity to see how a comunity organizes itself.
I have noticed the last few weeks that some of my older bug reports have been poked at by the devs in some form. Some have been merged, some have been addressed, some have been dismissed, some have been confirmed, and some have been marked as known issues.
It’s nice to see that the devs are becoming more organised so it is easier to follow their short and long term plans.
if they create a page that allows to easily monitor and access the various bugs, todo and know issues with data trend graphs divided by modules it would be very useful.
It won’t be in 2.82 though, so grab tonight’s buildbot build for 2.83 if you need it in production. I do like how long standing issues are being tackled more often now.
It would be nice to see some love on the outstanding IO issues, esp. FXB import and Alembic ones.
FBX camera import is still completely borked, and Alembic still hasn’t got motion vector support for motion blur rendering. Also, replacing a existing Alembix sequence in the scene is not impossible?
These are some of the reasons we still struggle to fully implement Blender in our workflow.
Stop being the next AD with all the new features, do a step back and fix things for a couple of point releases please…
Libraries are being upgraded to match the VFX reference platform 2020. (task )
The 2.82 Workboard was updated by Brecht with the missing documentations and critical bugs. Mantaflow bugs were not visited, but Sebastian Barschkis will tag them accordingly.
Reminder to developers to at the very least update the release log after committing a feature.
Aaron Carlisle (blendify) can help with the Reference Manual [8] update, but it is up to the developers to be sure Aaron has everything he needs to document the feature (help him with images, and pointing him where he can learn how to use the features. If not sure about this contact release manager - Nathan Letwory (jesterKing).
Ray (LazyDodo) asked for review of D6542 , which Brecht did during the meeting already.
Blender 2.83 - bcon1
Bastien Montagne brings the attention to T73268 . The developmenet team should investigate other ways to make complex GTests, than linking whole blend code over and over.
Dalai Felinto brings up that we did a really bad job in 2.82 of spreading new feautures throughout the bcon1 cycle. Let’s not let everyone to the last days again.
Changes and New Features
Hair physics collision was much improved, sharing code with the cloth solver. (video ) ( Luca Rood )
Python no longer uses environment variables like PYTHONPATH , unless --use-python-env is specified on the command line. This prevents hard to understand crashes when such environment variables point to incompatible Python versions. ( Campbell Barton )
Scene preview frame range is no longer clamped to the render frame range. ( Sebastian Parborg )
Extend selection with the Tweak tool is now more responsive, happening on mouse press. ( William Reynish )
I can understand this, but confining all user feedback to a section that’s generically called “Other Topics” does feel a bit like segregation, while Blender should be an open platform with constructive communication between developers and users.
the reason is that this user feedback is out of control. there is too much feedback that can’t be sorted and translated to actual development. They need a moderator who keeps contrusctive feedback open and keep the discussion going so papers and tasks can be created and who immediately closed spam threads.